R Notebook

signatures = read_csv("dados//signatures.csv")

Parsed with column specification:
cols(
  appliance = col_character(),
  state = col_character(),
  m1 = col_double(),
  m2 = col_double(),
  m3 = col_double(),
  m4 = col_double(),
  m5 = col_double(),
  m6 = col_double(),
  m7 = col_double(),
  m8 = col_double(),
  m9 = col_double(),
  m10 = col_double()
)

signatures$diff_1 = signatures$m2 - signatures$m1
signatures$diff_2 = signatures$m3 - signatures$m1
signatures$diff_3 = signatures$m4 - signatures$m1
signatures$diff_4 = signatures$m5 - signatures$m1
signatures$diff_5 = signatures$m6 - signatures$m1
signatures$diff_6 = signatures$m7 - signatures$m1
signatures$diff_7 = signatures$m8 - signatures$m1
signatures$diff_8 = signatures$m9 - signatures$m1
signatures$diff_9 = signatures$m10 - signatures$m1
signatures[3:7] %>% 
  gather(key = "variavel", value = "valor") %>%
  ggplot(aes(x = log(valor))) + 
    theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
    geom_histogram(fill = "lightyellow", color = "black", bins = 20) + 
    facet_grid(. ~ variavel, scales = "free_x")

signatures[8:12] %>% 
  gather(key = "variavel", value = "valor") %>%
  ggplot(aes(x = log(valor))) + 
    theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
    geom_histogram(fill = "lightyellow", color = "black", bins = 20) + 
    facet_grid(. ~ variavel, scales = "free_x")

signatures[13:17] %>% 
  gather(key = "variavel", value = "valor") %>%
  ggplot(aes(x = log(valor))) + 
    theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
    geom_histogram(fill = "lightyellow", color = "black", bins = 20) + 
    facet_grid(. ~ variavel, scales = "free_x")

signatures[18:21] %>% 
  gather(key = "variavel", value = "valor") %>%
  ggplot(aes(x = log(valor))) + 
    theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
    geom_histogram(fill = "lightyellow", color = "black", bins = 20) + 
    facet_grid(. ~ variavel, scales = "free_x")

# Determine number of clusters
wss <- (nrow(signatures[13:21]))*sum(apply(signatures[13:21],2,var))
for (i in 2:15) wss[i] <- sum(kmeans(signatures[13:21], 
    centers=i)$withinss)
plot(1:15, wss, type="b", xlab="Number of Clusters",
  ylab="Within groups sum of squares")

k_means_result <- kmeans(signatures[13:21], centers = 4, nstart = 20)
aggregate(signatures[13:21],by=list(k_means_result$cluster),FUN=mean)

grouped_signatures <- data.frame(signatures[13:21], k_means_result$cluster)
clusplot(grouped_signatures, k_means_result$cluster, color=TRUE, shade=TRUE, 
    labels=2, lines=0)

s

grouped_signatures$.row <- rownames(grouped_signatures)
  teste <- melt(grouped_signatures, id = c(".row", "k_means_result.cluster") )
parallel_plot = ggplot(teste, aes(x = variable, y = value, group = .row, colour = k_means_result.cluster)) + 
    theme(axis.text.x = element_text(angle = 45, hjust = 0)) +
    geom_line(alpha = .2) + 
    facet_wrap(~ k_means_result.cluster)
ggplotly(parallel_plot)

We recommend that you use the dev version of ggplot2 with `ggplotly()`
Install it with: `devtools::install_github('hadley/ggplot2')`